Dosimetric impact of MRI and PSMA-PET discordance in intraprostatic boost planning for high-risk prostate cancer

Partial-gland intraprostatic boost (IPB) during stereotactic body radiation therapy (SBRT) aims to improve tumor control while limiting toxicity. Whether magnetic resonance imaging (MRI) or prostate-specific membrane antigen positron emission tomography (PSMA-PET) provides the optimal target for focal dose escalation remains uncertain. This retrospective, single institution dosimetric study compared anatomic, volumetric, and dosimetric differences between MRI-based SBRT planning with IPB and PET-based segmentation. From April of 2023 to May of 2024, 20 consecutive patients with high-risk prostate adenocarcinoma were treated with MRI-based IPB. Thirteen patients received a two-fraction boost (19 Gy), and seven received a five-fraction regimen (36.25 Gy). PET scans were retrospectively fused with planning images, and PET-avid nodules were contoured. Volumetric and dosimetric variables were compared using Wilcoxon signed-rank test. Anatomic discordance was evaluated qualitatively and quantitively using dice coefficient and centroid distance thresholds. MRI-based segmentation identified nodules with a median interquartile range (IQR) volume of 1.45 (0.86–3.29) cm³, while PET-based nodules were larger at 2.17 (1.27–4.12) cm³ ( p = 0.04). The median Dice coefficient was 0.31 (range, 0–0.60), and the median centroid distance was 7.36 (range, 2.88–33.06) mm. Anatomic discordance occurred in 7 (35%) patients meeting the predefined criteria. MRI-based planning achieved a median D95% of 45.3 (range, 44.56–45.78) Gy for MRI-defined IPB in five fraction regimen versus 41.6 (range, 40.95–42.60) Gy for PET-defined nodules ( p < 0.001). Median D95% was 23.4 (range, 23.01–23.88) Gy versus 21.7 (range, 20.1–23.11) Gy for MRI and PET defines nodules, respectively, in the two-fraction boost regimen ( p < 0.001). In discordant cases, MRI-based planning delivered median (IQR) coverage of 125% (124–126%) for MRI-defined nodules compared to 115% (111–116%) for PET-defined nodules. These anatomic, volumetric, and dosimetric differences demonstrate lower dosimetric coverage of PSMA-PET-defined lesions when boost planning is based solely on MRI segmentation, supporting further investigation of combined-modality target delineation for SBRT focal boost planning. Not applicable.